r - 如何对 R 中的预测数据应用 wilcox 检验？

Question

当我们有一个数据文件时，我们使用以下代码进行 k 折交叉验证训练数据，

set.seed(308)

rand_search <- train(
    Effort ~ ., data = d,
    method = "svmRadial",
    ##Create 20 random parameter values
    tuneLength = 20,
    metric = "RMSE",
    preProc = c("center", "scale"),
    trControl = rand_ctrl
) 
  model1 <- predict(rand_search, newdata = test1)

And another search algorithm like grid
grid_search <- train(
    Effort ~ ., data = d,
    method = "svmRadial",
    ##Create 20 random parameter values
    tuneLength = 20,
    metric = "RMSE",
    preProc = c("center", "scale"),
    trControl = rand_ctrl
) 
model2 <- predict(grid_search, newdata = test1)

我的问题是，如果我们必须找到显着性检验（wilcox 检验），我们该如何应用它？我们是否需要像下面这样通过 mode1 和 model 2 进行 wilcox 测试？

wilcox.test（模型1，模型2）

Answer 1

在trainControl您不需要指定数据。在train函数中，您必须提及数据，例如

#Model training
set.seed(308) 
rand_search <- train(Effort ~ ., data = train1 ,
                                method = "svmRadial",
                                ## Create 20 random parameter values
                                tuneLength = 20,
                                metric = "RMSE",
                                preProc = c("center", "scale"),
                                trControl = rand_ctrl)

并且test1应该用于预测目的，例如

#For calibration
models_cal <- predict(rand_search, newdata = train1)
#For independent validation
models_val <- predict(rand_search, newdata = test1)